Lemma 45.9.5. Assume given (D0), (D1), (D2), and (D3) satisfying (A), (B), and (C). Let X be a smooth projective scheme over k. If X = \emptyset , then H^*(X) = 0. If X is nonempty, then \gamma ([X]) = 1 and 1 \not= 0 in H^0(X).
Proof. First assume X is nonempty. Observe that [X] is the pullback of [\mathop{\mathrm{Spec}}(k)] by the structure morphism p : X \to \mathop{\mathrm{Spec}}(k). Hence we get \gamma ([X]) = 1 by axiom (C)(a) and Lemma 45.9.4. Let X' \subset X be an irreducible component. By functoriality it suffices to show 1 \not= 0 in H^0(X'). Thus we may and do assume X is irreducible, and in particular nonempty and equidimensional, say of dimension d. To see that 1 \not= 0 it suffices to show that H^*(X) is nonzero.
Let x \in X be a closed point whose residue field k' is separable over k, see Varieties, Lemma 33.25.6. Let i : \mathop{\mathrm{Spec}}(k') \to X be the inclusion morphism. Denote p : X \to \mathop{\mathrm{Spec}}(k) is the structure morphism. Observe that p_*i_*[\mathop{\mathrm{Spec}}(k')] = [k' : k][\mathop{\mathrm{Spec}}(k)] in \mathop{\mathrm{CH}}\nolimits _0(\mathop{\mathrm{Spec}}(k)). Using axiom (C)(b) twice and Lemma 45.9.4 we conclude that
is nonzero. Thus i_*\gamma ([\mathop{\mathrm{Spec}}(k)]) \in H^{2d}(X)(d) is nonzero (because it maps to something nonzero via p_*). This concludes the proof in case X is nonempty.
Finally, we consider the case of the empty scheme. Axiom (B)(a) gives H^*(\emptyset ) \otimes H^*(\emptyset ) = H^*(\emptyset ) and we get that H^*(\emptyset ) is either zero or 1-dimensional in degree 0. Then axiom (B)(a) again shows that H^*(\emptyset ) \otimes H^*(X) = H^*(\emptyset ) for all smooth projective schemes X over k. Using axiom (A)(b) and the nonvanishing of H^0(X) we've seen above we find that H^*(X) is nonzero in at least two degrees if \dim (X) > 0. This then forces H^*(\emptyset ) to be zero. \square
Comments (0)